Re: A few bonobo & chimp genes more related to our own

Ernest Major wrote:
In message <d15eb192-6019-413f-9663-2b02e78e6811@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>, VoiceOfReason <papa_fox@xxxxxxxxxxxxx> writes


Ron O wrote:
On Apr 20, 5:46pm, Glend <interelectromagne...@xxxxxxxxx> wrote:
> "But when David Reich, a population geneticist at the Broad Institute
> in Boston, and his colleagues sequenced tens of thousands of random
> snippets of DNA from bonobos and chimps, then compared these to the
> matching sequences of the complete chimp and human genomes, they found
> a surprise. In 453 of their sequences, either bonobos or chimps proved
> to be more closely related to humans than to each other."
>
> http://www.newscientist.com/article/dn13734-ghost-of-genetics-past-sh...
>
> Actually, I don't think it's much of a surprise. Why wouldn't some
> genes be relatively conserved in the bonobo or the chimp line, while
> diverging so much in the other that the conserved genes would be more
> related to ours than to the ones that have evolved rapidly since
> bonobos and chimps split?
>
> But it's interesting as more of the possibilities afforded by actual
> science (evolution), and so much meaningless data to the IDists.
>
> Certainly, it's another bit that seems to me to be expected (so long
> as it's few, which 453 is) under evolution by known mechanisms, while
> having no sense by design considerations. Then again, what does make
> sense in "design" of life?
>
> Glen Davidsonhttp://tinyurl.com/2kxyc7

What did the paper say? It isn't unexpected, but I don't know the
distribution. 453 may have been less than 5% (you are talking about
the sequences of three closely related species). In coding sequences
there is only around 1% difference between humans and both chimps, and
the two chimps have been evolving separately for around half that
time. You are going to have quite a lot of sequence that doesn't
change at all. They have to differentiate coding from non coding and
then you have to figure on sequence length. Shorter sequences are
more likely to have no or fewer changes. Just think if there is only
a 1% difference you are going to have a heck of a lot of sequences
that are identical that are around 1000 base-pairs in length in a
genome of 3 billion base-pairs just by chance. 1000 is about max for
Sanger sequencing and the new generation sequencers produce sequences
that are even shorter.

Help me Landru, help me!

Let's pretend for a moment that I know diddly about the details of
genetics (which, by the way, is not a difficult supposition to
simulate...)

After reading some of the other comments, I'm left with the
question... Does this mean that chimps and bonobos are (likely) more
closely related to humans than they are to each other?

No. Apart from the evidence to the contrary, I don't that that is possible topology - at least in the absence of reticulation.

Or that
there's some preliminary evidence that suggests that they are or might
be?

No. You should notice that the paper refers to 453 sequences among tens of thousands.


Or am I totally misconstruing the article in question?


What you get from homologous stretches of DNA from different species is a gene tree. This is usually pretty similar to the underlying species tree, but there are a variety of processes which can cause them to diverge.

I was about to write you a primer on how gene trees can diverge from species trees, and what can be done (more data!) to solve the problem, but I now realise that would take me some time to organise my knowledge on this point - it's not something I've thought all that closely about.

Perhaps John Harshman will fill the gap. (POTM opportunity :-)).

I've explained the most likely answer for the current case in another post. But here's a classification of sorts:

I. The gene tree has been incorrectly determined: that is, phylogenetic analysis has produced an estimated tree that doesn't match the true gene tree. This is certainly possible, and not all that uncommon, but in the current context it's boring, and I'll ignore it.

II. The gene tree is correct and really doesn't match the species tree.

A. Horizontal transfer.

1. Hybridization.

2. Genuine leakage from one genome to another, e.g. through retroviral insertion.

B. Paralogy. A gene has been duplicated and you are looking at different copies in the various species. It may be that some copies have gone extinct, making it easier to sequence different copies by mistake.

C. Lineage sorting. Different alleles existed in the ancestral population, polymorphism was maintained through at least two speciation events, and has randomly sorted out in a way contrary to the species phylogeny. (That's the most likely in the chimp/bonobo/human and chimp/human/gorilla cases.) A common metaphor for this is to imagine the species phylogeny as a pipe whose width depends on population size, and gene lineages as lines rattling around in the pipe, splitting and disappearing with a frequency depending on that population size.

That's what I think of now. For many of them, the cure is to sequence a lot of loci. If you get a strong majority vote, that should be the species phylogeny. If you don't get such a vote, things can be more difficult. If you get a 50-50 split, hybridization is a good guess, and you should be able to recover the phylogenies of both separate lineages.

In the case of paralogy, you may with greater effort be able to recover both paralogs from some or all of the species, which should give you two species trees that should match.

In the case of lineage sorting, the absence of a strong majority vote would suggest that the situation is approaching a polytomy. The larger the ancestral population and the shorter the time between speciations, the more likely lineages are to sort randomly. At some point we have to give up the idea of a true, bifurcating phylogeny and accept that a species has split into three. To make matters a bit worse, the probablities of getting different trees by random sorting are not equal; there is a bias toward trees with a symmetrical shape, and this may produce a false convergence on a wrong species phylogeny. The cure for that is to model evolution to take this sorting into account.

.

Relevant Pages

  • Re: Phylogenetic question for John Harshman
    ... Gene Poole wrote: ... by showing you can make such a tree from ... genes can be introduced in other species. ... sequences that make a different conclusion ...
    (talk.origins)
  • Re: Phylogenetic question for John Harshman
    ... John Harshman wrote: ... by showing you can make such a tree from ... genes can be introduced in other species. ... sequences that make a different conclusion ...
    (talk.origins)
  • Re: Cambrian explosion bit of an embarassment
    ... when sequences from some species in a phylogenetic analysis have ... making them long branch species. ... used as an outgroup to root the tree inevitably has a relatively long ... trees from the characters (DNA sequences or whatever) of the various ...
    (talk.origins)
  • Re: Cambrian explosion bit of an embarassment
    ... "The pernicious effects of long branch attraction occur ... making them long branch species. ... used as an outgroup to root the tree inevitably has a relatively long ... trees from the characters (DNA sequences or whatever) of the various ...
    (talk.origins)
  • Re: Part 1 (of 3): What are major aspects of evolutionary theory?
    ... But if I have an unrooted tree, ... > branching node adjacent to Harrietphyta is only around 65-70 million ... > extremely long-lived species which is ancestral to the branch point ...
    (talk.origins)
Loading